A Neuro-Fuzzy Approach for the Fault Location Estimation of Unsynchronized Two-Terminal Transmission...

7/29/2019 A Neuro-Fuzzy Approach for the Fault Location Estimation of Unsynchronized Two-Terminal Transmission Lines

1/15

International Journal of Computer Science & Information Technology (IJCSIT) Vol 5, No 1, February 2013

DOI : 10.5121/ijcsit.2013.5102 23

A NEURO-FUZZYAPPROACH FOR THE FAULT

LOCATION ESTIMATION OF UNSYNCHRONIZED

TWO-TERMINALTRANSMISSION LINES

Ramadoni Syahputra

Department of Electrical Engineering, Faculty of Engineering,

Universitas Muhammadiyah Yogyakarta, Yogyakarta, 55183, [email protected]

ABSTRACT

Overhead transmission line is an element of electrical power systems that are most frequently experiencedshort circuit faults compared to other power system elements. Short circuit faults on overhead transmission

line cause a relatively large current and therefore can damage mechanically the electrical equipment

connected to the system. The protection system is essentially needed in this situation. In addition, it takes a

piece of equipment that can detect the location of fault in order to expedite the repair process, especially if

the fault is permanent state. In this paper, a neuro-fuzzy aprroach for short circuit fault location estimation

which uses data from both ends of overhead transmission line is described. The approach utilizes the

advantages of digital relaying which are available today. The unsynchronized data of fault voltages and

currents at two-end of overhead transmission line is applied in this technique. The accurate fault location

estimation technique has irrespective of source impedances, fault resistances, fault types, and load

currents. Simulation of short circuit fault of transmission line has done by using EDSA software. Short

circuit currents and voltages from both ends of overhead transmission line have used to input data of

neuro-fuzzy method in Matlab program. Simulation results demonstrate the accuracy of the method. The

results shows that the lowest estimation error for single phase to ground fault with the variation of fault

resistances of 0 ohms, 10 ohms, 50 ohms, and 100 ohms, respectively, is 0.0027%, while the highestestimation error is 0.2962%.

KEYWORDS

Neuro-Fuzzy, Two-terminal fault location algorithm, transmission line, unsynchronized sampling.

1. INTRODUCTION

The transmission line is an element of electrical power systems that are most frequently

experienced short circuit faults compared to other electrical power system elements, with thepercentage of probability of occurrence of 50% [1]. There are four types of short circuit faults on

a transmission line, i.e., single phase to ground fault, phase to phase fault, double phase to goundfault and symmetrical three-phase fault. Short circuit faults on overhead transmission line cause arelatively large current and therefore can damage mechanically the electrical equipment

connected to the system. In addition, it takes a piece of equipment that can detect the location offault in order to expedite the repair process, especially if the fault is permanent. The continuous

and reliable electrical energy supply is the aim of electrical power system operation. The shortcircuit faults on the line must be estimated accurately to allow maintenance operators to arrive at

the scene and repair the faulted transmission line part as soon as possible. Geographical layout

and rugged terrain make some sections of electrical power transmission lines difficult to reach;
mailto:[email protected]:[email protected]


2/15


24

therefore, the robustness of fault location estimation under a variety of power system operating

constraints and fault conditions is an important requirement. Therefore, it is important to knowwhere the location of a faulted transmission line section has been occured. Development of the

application of digital and microprocessor-based power protection systems has been motivated forfault location technique researchers in the last few decades [2]-[4]. However, accuracy of the

distance to the fault determination from a power substation is affected by several stochasticfactors. There are some factors which affecting the accuracy of fault estimation, i.e., faultresistance, equivalent impedances of transmission line, effect of the load variation and

imprecisions of the transmission line parameters measurements [4]-[6]. Therefore, the accuracy offault distance determination from both ends of transmission line may be insufficient.

Accurate location of faults on overhead electrical power transmission lines for the inspection-repair purpose is of vital importance for utility staff and operators for expediting service

restoration, and thus to reduce outage time, operating costs and customer complains. During thelast decade a number of fault location algorithms have been developed, including the transient-

state approach, steady-state phasor approach, the traveling-wave approach, the lumped parameter

approach, and the differential equation approach [7], as well as two-end [8] and one-end [9]algorithms. In the last category, synchronized [10] and non-synchronized [11] sampling

techniques are used. However, two-terminal data are not widely available on the protectionequipment. Fault location determination methods using the measured voltage and current at oneterminal channel is still inadequate because it does not take into account the current channel

interference from the other terminal. In [12] usage of synchronized measurements of currents andvoltages from all two terminals of transmission line has been considered. They have used the

distributed parameter models of the overhead transmission line sections. The approach assureshigh accuracy of fault location estimation, and the faulted transmission line section is also rely on

indicated [12]. Then the use of three-terminal unsynchronized measurements of short circuit faultcurrents and voltages has been considered in [13]. They have used the lumped parameter models

of the overhead transmission line sections. It was implied that the error estimation resulting from

such simplification is minimized due to the redundancy of the fault location equations. Yetanother utilization of three end unsynchronized measurements has been proposed in [14], where

exchanging the minimal amount of information between the line terminals over a protection

channel was considered. After that, the development of fault location techniques for threeterminal transmission lines that utilizing only from two-end synchronized measurements of

currents and voltages has been studied in [15]-[16]. From a practical viewpoint, it is desirable forequipment to use only one-terminal data. The one-end algorithms, in turn, utilize different

assumptions to replace the remote end measurements. Most of fault location algorithms are onlybased on local measurements. The use of fuzzy logic based in electrical power system study hasbeen became an interesting research in two last decades [17]. Currently, the most widely used

method of overhead transmission line fault location is to determine the apparent reactance of the

line during the time that the short circuit fault current is flowing and to convert the ohmic result

into a distance based on the parameters of the line. It is widely recognized that this method issubject to errors when the fault resistance of the transmission line is high and the line is fed fromboth ends (substation), and when parallel circuits exist over only parts of the length of the faulty

line.

In this paper, a neuro-fuzzy approach for short circuit fault location estimation, which utilizesunsynchronized measurements of currents and voltages at two-terminal of transmission line isproposed. The measurements has used to input data for training process in neuro-fuzzy system.

The lumped parameter in medium transmission line model is strictly used. The approach can beapplied in digital distance protection which is available today in electrical power system. The

method allows for accurate estimation of short circuit fault location irrespective of fault types,


3/15


25

fault resistance, load currents, and source impedances. The powerful of both EDSA software and

Matlab software have used in this study.

2. FUNDAMENTAL THEORY

2.1. Faults in Transmission Lines

Transmission lines is one of the main component in electric power systems. Transmission line is acomponent of electrical power systems that are most frequently experienced faults compared to

other power system elements. Faults in power systems are basically classified as shunt faults andseries faults [18]. Both types can be balanced fault or unbalanced fault. Faults on a transmissionline can occur as whether single or simultaneous faults. Most of faults in transmission lines are

single line to ground type. Therefore, the single line to ground fault with various fault resistancesis used in this study.

Simultaneous faults on a transmission line consist of a combination of the same or different types

of faults. The preferred model for fault calculation is the nodal approach in the frequency domain

with transmission line symmetrical components or in the time domain with space phasors or itscomponents. Conventional techniques for fault location calculation are usually based on

admittance equations in admittance form. Another general problem is the treatment ofsimultaneous line faults. Each fault is characterized by the so-called boundary conditions

regarding the currents and voltages at the fault location of transmission line.

In an electrical power system comprising of any various interacting elements, there always exists

a possibility and probability of faults. The emerge of large power generating stations and highlyinterconnected power systems via any overhead transmission lines makes early fault detection

and rapid equipment isolation imperative to maintain the stability of the system. Faults onoverhead transmission lines need to be detected rapidly, located accurately and repaired as soon

as possible. Fault detector module of a transmission line protective device can be used to start

other relaying modules. The detectors provide an additional way of security in a power protection

relaying application as well, and the location of short circuit fault must be determined.

Besides being used to accurately locate a fault, such an method can be used for automated faultanalysis. Any occurrence of a short circuit fault should be detected and cleared by the power

protective relaying devices. A power protective relaying operation analysis is required if anassessment of its performance is needed. In order to perform the analysis, one has to have a

reference method with which to compare the digital relay operation. The fault location techniquethat can provide both fault type classification and accurate location is an ideal reference for the

robust digital protective relaying operation. The technique can be incorporated into a fault

analysis automatically by providing high speed information of the fault type and fault location.This is important information for determining if a power protective relay has operated correctly

since the relay is also supposed to determine both type and location of fault. The location of faultdetermined by the distance relay does not have to be too accurate since it only has to determine

the zone of the occurrence of fault. The appropriate location of the fault provided by the faultlocation technique is more accurate and is needed by power operators.

2.2. Neuro-Fuzzy Method

During the last two decades adaptive neuro-fuzzy approach has been became a popular method incontrol area. In this part, a brief description of the adaptive neuro-fuzzy inference system

(ANFIS) principles is given which are refered to [19]. The fundamental structure of the type of


4/15


26

fuzzy inference system (FIS) could be seen as a model that maps input characteristics to input

membership functions. After that, it maps all membership function as input to rules and rules to aset of characteristics of FIS output. On the last step, FIS maps characteristics of output to

membership functions as output, and the membership function as output to a decision associatedwith the output. As can be seen that FIS has been stated only non-arbitrary membership functions

that were chosen arbitrarily. Fuzzy inference system (FIS) is only used to modeling systemswhose the structure of fuzzy rule is essentially predetermined by the operator interpretation of thevariable characteristics in the model. However, it cannot be distinguish what the FIS membership

functions should look like simply from the data for some situations. Parameters of FIS could bechosen so as to tailor the membership functions to the input and output data rather than choosingthe parameters associated with a given membership function arbitrarily in order to account for

these types of variations in the values of data. Therefore, the necessity of an adaptive properties infuzzy inference system becomes obvious.

The adaptive neuro learning concept works similarly to the artificial neural networks. Neuro-

adaptive learning techniques provide a method for the fuzzy modeling procedure to learn

information about a data set. It computes the membership function parameters that best allow theassociated fuzzy inference system to track the given input and output data. A network-type

structure similar to that of an artificial neural network can be used to interpret the input andoutput map so it maps inputs through input membership functions and associated parameters, andthen through output membership functions and associated parameters to outputs. Through the

learning procedure, parameters which associated with the membership functions will changes.The parameters computation is facilitated by a vector of gradient. The vector of gradient

determine a criterion of how well the fuzzy inference system (FIS) is modeling the input andoutput data for a given parameters set. While the vector of gradient is obtained, several

optimization procedures can be used in order to control the parameters to reduce some errormeasure (index of performance). The error measure is comonly defined by the sum of the squared

difference between desired and actual outputs. ANFIS uses a combination of back propagation

procedure and least squares estimation for membership function parameter estimation.


5/15


27

Figure 1. Sugenos fuzzy logic model

Figure 2. The architecture of the ANFIS.

The suggested ANFIS has several properties:

1. The ANFIS output is Sugeno-type of zero-th order.

2. ANFIS has only a single output which obtained using defuzzification process of weightedaverage. All output membership functions are constant.

3. It has no rule sharing. The number of rules must be equal to the number of output

membership functions.

4. It has unity weight for each rule.

A1

A2

B1

B2

x

y

N

N

f

x y

x y

2w

1w

w1

w2

11fw

22fw

Layer 1 Layer 2 Layer 3 Layer 4 Layer 5

A1

A2

B1

B2

x

x

y

y

w1

w2

f1 = p1x + q1y + r1

f2 = p2x + q2y + r2

21

2211

ww

fwfwf

+

+=


6/15


28

Figure 1 shows Sugenos fuzzy logic model. The ANFIS architecture is shown in Figure 2. The

architecture comprising by input, fuzzification layers, inference unit and defuzzification layers.The ANFIS architecture can be desripted as consisting of N neurons in the input layer and F

membership functions for each input, and F*N neurons in the fuzzification layer. The inferenceunit and defuzzification have FN rules with FN neurons, while the output layer has one neuron.

For simplicity, it is assumed that the fuzzy inference system under consideration has two inputs xand y and one output z as shown in Figure 2. For a zero-order Sugeno fuzzy model, a commonrule set with two fuzzy if-then rules is the following:

Rule 1: If x is A1 and y is B1, Then f1 = r1 (1)

Rule 2: If x is A2 and y is B2, Then f2 = r2 (2)

Here the output of the i-th node in layer n is denoted as On,i:

Layer 1. Every node i in this layer is a square node with a node function:

1

iO = Ai(x), for i = 1, 2, (3)

or,1

iO = Bi-2(y), for i = 3, 4 (4)

wherex is the node-i input, andAi is the label of linguistic terms (big, low, etc.) associated with

this node function.1

iO is the Ai membership function.

1

iO specifies the degree to which the given

x satisfies the Ai. Usually Ai(x) is chosen to be bell-shaped with maximum equal to 1 andminimum equal to 0, such as the generalized bell function:

i

i

i

A 2b

a

cx1

1(x )

+

= (5)

Parameters in this layer are referred to aspremise parameters.

Layer 2. Each node in layer 2 is labeled by which multiplies the incoming data and sends the

product out. For instance,

2

iO = wi = Ai(x) x B(y), i = 1, 2. (6)

Each node output represents the firing strength of a rule. Other T-norm operators which showsgeneralized AND can be used in layer 2.

Layer 3. Every node in this layer is a circle node labeled N. The i-th node calculates the ratio of

the i-th rules firing strength to the sum of all rules firing strengths:

21

i3

iww

wwO

+

== , i = 1, 2. (7)

The outputs of layer 3 will be mentioned as normalized firing strengths.

Layer 4. Every node i in this layer is a square node with a node function:


7/15


29

)iiiiii4

i ryqx(pwfwO ++== (8)

iw is the layer 3 output, while {pi, q i, r i} is the set of parameter. All parameter in this layer will

be mentioned as consequent parameters.

Layer 5. The single node in this layer is a circle node labeled that computes the overall outputas the summation of all incoming signals, i.e.,

= ii5i fwO (9)

2.3. Unsynchronized Sampling

The procedure of short circuit fault location estimation in this study is use two terminals ofelectrical power transmission line as shown in Figure 3 [11].

VA

VB

IA

IB

BUS A BUS B

If

Vf

VA

E VB

E

ZA

ZBmZ (1-m)Z

Figure 3. Short circuit fault in transmission lines.

Transmission line as shown in Figure 3, both voltage and current phasors from protected two

terminals of the line are required in this procedure, but unsynchronized. As can be seen in Figure3, the method will estimate fault distance m from two ends of overhead electrical powertransmission line.

The fault voltages and currents from bus A and bus B of transmission line are not synchronized,

while is angle synchronization. For example, the voltage at bus A and bus B can be written asfollow:

VA = |VA| m+; VB = |VB| m (10)

where m and m are measured angle from two ends respectively, and is the synchronizationphasor angle between bus A and bus B. The similar equation can be written for current phasors.Hence, equation (10) has become:

VA ej

VB + Z IB = mZ (IA ej

+ IB) (11)

The unknown components of equation (11) are distance fault m and complex number = ej

.Equation (11) can be separated into real part and imaginary part to forms the new two equationsas follow:

Re(VA)sin+ Im(VA)cos- Im(VB) + K4 = m(K1 sin+ K2 cos+ K4) (12)

Re(VA)cos - Im(VA)sin- Re(VB) + K3 = m(K1 cos - K2 sin+ K3) (13)


8/15


30

Coefficients of K1, K2, K3, and K4 in equation (12) and equation (13) can be defined as follow:

K1 = R Re(IA) X Im(IA) (14)

K2 = R Im(IA) + X Re(IA) (15)

K3 = R Re(IB) X Im(IB) (16)

K4 = R Im(IB) + X Re(IB) (17)

Then, the equation with unknown angle is formed. As rearranged the equations above, then thenew equations are resulted as follow:

a sin + b cos + c = 0 (18)

where,

a= K3Re(VA) K4Im(VA)-K1Re(VB) K2Im(VB)+K1K3+K2K4 (19)

b= K4Re(VA)K3Im(VA)K2Re(VB) +K1Im(VB)+K2K3 K1K4 (20)

c= K2Re(VA)-K1Im(VA)-K4Re(VB)+K3Im(VB) (21)

From the equation (18) can be seen that angle (synchronization angle) is unknown. Theunknown one can be found by using Newton-Raphson iterative algorithm. The equation for

iteratively to count the angle (in radian) is:

)('

)(1

k

k

kkF

F

=

+(22)

Iterative process will be stopped when the difference between two end values that smaller thanthe stated float is achieved, for example: 4

1 10

+


9/15


31

And, error estimation can be calculated by the equation below [19]:

%100(%)

=

lengthline

locationestimatedlocationactualestimationError (25)

3. SIMULATION RESULTS

The procedure of this research is shown in Figure 4. Simulation of short circuit fault of

transmission line has done by using EDSA software. Short circuit currents and voltages from bothends of overhead transmission line have used to input data of neuro-fuzzy method in Matlab

environment. It is used as the main engineering tool for performing modelling and simulation ofelectric power systems, as well as for interfacing the user and appropriate simulation programs.

MATLAB has been chosen due to availability of the powerful set of programming tools, signalprocessing, numerical functions, and convenient user-friendly interface. In this specially

developed simulation environment, the evaluation procedures can be easily performed. We have

used Fuzzy logic Toolbox of MATLAB to develop the ANFIS model with 2 inputs and singleoutput as given in Figure 2. The simulations were performed by an Intel Pentium core(TM) 2

duo CPU, 1.80 GHz, 4 GB RAM.

For evaluating the performance of the proposed algorithm, the author adopts EDSA software for

fault data generation and Matlab for neuro-fuzzy algorithm implementation. Power system shownin Figure 3 has selected for the studies reported in this paper. Bus A and bus B were considered to

be connected by 150 kms, 500 kV transmission lines [21]. Two equivalent power systemsconsidered to be connected to bus A and bus B. Electrical parameters of transmission lines andequivalent power sources are given in Table I.

TABLE I

ELECTRICAL PARAMETERS OF TRANSMISSION LINE IN THE STUDY

Components

Impedances

Positive sequence(Z1)

Negative sequence(Z2)

Zero sequence(Z0)

Transmission lines 53,09086,07 53,09086,07 173,68372,96Power source A 4,00189,03 3,50188,90 1,41445Power source B 100,02088,85 75,02788,47 25,07285,65


10/15


32

Figure 4. Procedure of the research.

In order to estimate the fault location of transmission line accurately, the fuzzy systems aretrained with a separate ANFIS structure and suitable off-line data. The main steps of the

procedure are (see Fig. 4):

Step 1: Create the architecture of fuzzy inference system (FIS) in Matlab environment. Thearchitecture consist of two inputs (i.e. voltages and currents fault measured in the locator

end of transmission line) and one output.Step 2: Determine the membership functions for ANFIS input and output, recpectively, and then

define the If-Then rules. In this work, gbell membership function has been choosen for

each input and output of FIS architecture.Step 3: Collecting or producing suitable information (data) to train ANFIS. The data for training

process should have same form and the various conditions of a real power systemis

included. A power system simulation using EDSA has been carried for achieving thesuitable data.

Start

Literature study

Create the transmission line model in EDSA

Create the neuro-fuzzy fault locator in Matlab

Examine the short circuit fault on transmission line

Use the fault currents and voltages as input data of

neuro-fuzzy program in Matlab

Examine the

other faults?Yes

No

Analyze the performance neuro-fuzzy method

Conclusion

Finish

Examine neuro-fuzzy program

for fault type and fault location estimation


11/15


33

Step 4: Training process. The suitable data which are collected in step (3) are presented to

network and adaptive nodes are adjusted. This presses will be stopped when error meetproposed goal. The nodes of adaptive properties update after entire patterns have been

presented to network.Step 5: Testing process. In training procedure, the fault locator should be given an acceptable

output for unseen data. If test pattern output reached an acceptable range, fuzzy rule isadjusted in the best conditions.

Figure 5 shows the membership function of input variable Voltage, while Figure 6 shows themembership function of input variable Current. The voltage and current input variables haveaddressed as the the fault variables of transmission line under consideration. Figure 7 shows the

training data of ANFIS and the ANFIS output for 20 epochs.

For pre-fault calculations, all transmission lines were modelled by equivalent pi networks and allloads were considered to be constant power loads. For calculating fault currents and voltages on

the inception of a fault, the selected line has modelled by two equivalent pi networks, one for the

section from bus A to the fault and the other for the section from bus B to the fault. Forexamining the fault distance technique, it was assumed that digital distance relays have been

provided at the line terminals on bus A and bus B. It was also assumed that these digital distancerelays measure fundamental frequency voltages and currents from sampled data.

Figure 5. Membership function of input variable Voltage .

Figure 6. Membership function of input variable Current .


12/15


34

0 1 2 3 4 5 6 7 8 9 10

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Training Data

ANFIS Output

Figure 7. Training data of ANFIS and the ANFIS output.

TABLE II

FAULT DISTANCE ESTIMATION FOR SINGLE PHASE-TO-GROUND FAULT

Actual Distance

(km)

Estimation Error (%)

Rf = 0 Rf= 10 Rf= 50 Rf = 100

0 0,0027183 0,0024771 0,0093352 0,0170958

15 0,0125956 0,0033028 0,0119279 0,0219511

30 0,0090500 0,0080960 0,0221115 0,0390903

45 0,0195715 0,0163435 0,0358773 0,0600260

60 0,0211366 0,0280220 0,0532048 0,0847273

75 0,0227297 0,0431110 0,0740820 0,1131887

90 0,0243211 0,0615933 0,0985028 0,1454183

105 0,0258992 0,0834552 0,1264669 0,1814347

120 0,0274970 0,1086872 0,1579806 0,2212671

135 0,0291193 0,1372857 0,1930578 0,2649561

150 0,0307341 0,1582224 0,2184332 0,2962510

In order to test the powerful of neuro-fuzzy method in this research, simulation of a single phaseto ground fault has done in EDSA environment. The fault has occured on the selected location of

transmission line. Buses A and B and some locations, i.e. 15 kms, 30 kms, 45 kms, 60 kms, 75kms, 90 kms, 105 kms, 120 kms, and 135 kms, were chosen as fault locations. Fault resistanceswere varied from 0 ohms, 10 ohms, 50 ohms, to 100 ohms. Fundamental frequency voltages at

bus A and bus B and line currents were calculated and provided to the fault location program in


13/15


35

Matlab as inputs. Fault location of a transmission line studies for single phase to ground fault are

reported in this paper, as shown in Table II and Figure 8.

0

15

30

45

60

75

90

105

120

135

150

0

0.2

0.4

0.6

0.8

1

1.2

Estimationerror(%

Fault distance (km)

Estimation Error as a Function of fault Distance

for Single Phase to Ground Fault

Rf = 0 ohm Rf = 10 ohms

Rf = 30 ohms Rf = 50 ohms

Rf = 70 ohms Rf = 100 ohms

Rf = 10 ohms by Sachdev & Agarw al

Figure 8. Estimation error as a function of fault distance for single phase to ground fault on transmission

line.

Table II lists the estimation error of fault locations for a single phase to ground fault with faultresistances of 0 ohms, 10 ohms, 50 ohms, and 100 ohms, respectively. The estimation errorsexpressed as percentages of the line length are shown in Figure 5. The results indicate that

distances of faults estimated by the proposed method are substantially more accurate than thedistances estimated by Sachdev and Agarwal [22]. When a single phase to ground fault occurs in

bus A (0 kms distance) with fault resistance of 0 ohms, the estimation error is 0.0027 %. This

error value is the smallest estimation error in the study. As can be seen in Table II and Figure 8that the highest short circuit fault estimation error for single phase to ground fault is 0.2962% at

distance of 150 kms with fault resistance of 100 ohms.

4. CONCLUSIONS

This paper has proposed a neuro-fuzzy approach that estimates the distance of a transmission lineshort circuit fault from relay locations using unsynchronized fundamental frequency voltages andcurrents measured at the two ends of transmission line. In this paper, a neuro-fuzzy aprroach for

short circuit fault location estimation which uses data from both ends of overhead transmissionline is described. The approach utilizes the advantages of digital relaying which are available

today. The accurate fault location estimation algorithm has irrespective of source impedances,fault resistances, fault types, and load currents. Simulation of short circuit fault of transmission


14/15


36

line has done by using EDSA software. Short circuit currents and voltages from both ends of

overhead transmission line have used to input data of neuro-fuzzy method in Matlab program.Simulation results demonstrate the accuracy of the method. The results shows that the lowest

estimation error for single phase to ground fault with the variation of fault resistances of 0 ohms,10 ohms, 50 ohms, and 100 ohms, respectively, is 0.0027%, while the highest estimation error is

0.2962%.

ACKNOWLEDGEMENTS

The author would like to thank profusely and the highest appreciation for DIKTI (the Directorate

General of Higher Education) Ministry of Education and Cultural Affairs, Republic of Indonesia,for having funded this research.

REFERENCES

[1] Ram, B. and Vishwakarma, DN, "Power System Protection & Switchgear", pp. 3-6, McGraw-Hill

Pub. Co. Ltd., New Delhi, 1995.

[2] A.M. Borbely and J.F. Kreider, Distributed Generation: The Power Paradigm for the NewMillennium, CRC Press, Washington D.C., 2001.

[3] T. Takagi, Y. Yamakoshi, M. Yamuaura, R. Kondow, and T. Matsushima, Development of a New

Type of Fault Locator Using One Terminal Voltage and Current Data, IEEE Trans. On Power

Apparatus and System, vol. PAS-101, No 8, pp. 2892-2898, Aug. 1982.

[4] Eriksson, L., S.D. Rockfeller and M. Saha, Accurate Fault Loc. with Compensation for Apparent

Reac. in the Fault Resistance Resulting from Remote End in Feed, IEEE Transactions on Power

Aparatur and System, PAS-104, No 2, 1985.

[5] D. Novosel, D.G. Hart, M.M. Saha, and S. Gress, Optimal fault location for transmission system,

ABB Review 8, pp. 20-27, 1994.

[6] A. Sauhats and M. Danilova, Fault Location Algorithms for Super High Voltage Power

Transmission Lines, in Proc. Of IEEE Bologna Power Tech Conf., 2003.

[7] A. Sauhats and M. Bockarjova, Algorithms, Means and Tools of Fault Location on Transmission

Lines, Proc. EPE-PEMC'2004, Riga, Latvia, 2004.

[8] Lian, B and M.M.A. Salama, An Overview of Digital Fault Location Alg. For Transmission LineUsing Transient Waveforms, Electric Power System Research, Vol. 29, No. 1, pp. 1725, 1994.

[9] L.B. Sheng and S. Elangovan, A fault location algorithm for transmission lines, Electric Machines

& Power Syst., Vol. 26, No. 10, pp. 9911005, 1998.

[10] Q. Zhang, Y. Zhang, W. Song, and Y. Yu, Transmission line fault location for phase-to-earth fault

using oneterminal data, IEE Proc. Trans. Distribution., Vol. 146, No. 2, pp. 121124, 1999.

[11] Kezunovic, M and J. Mrkic, An Accurate Fault Location Alg. Using Sync. Sampling, Electric

Power System Research, Vol. 29, No. 3, pp. 161169, 1994.

[12] Novosel, D., E. Udren, J. Garitty and D.G. Hart, Unsynchronized Two Terminal Fault Location

Estimation, IEEE Transactions on Power Delivery, Vol. 11, No. 1, pp. 130138, 1996.

[13] Aggarwal, R.K., A.T. Johns, D.V. Coury and A. Kalam, A Practical Approach to Accurate Fault

Location on EHV Teed Feeders, IEEE Transactions on Power Delivery, Vol. 8, pp. 874883, July

1993.

[14] A.A. Girgis, D.G. Hart, and W.L. Peterson, A new fault location technique for two-and three-

terminal lines, IEEE Trans. Power Delivery, Vol. 7, No.1, pp. 98107, January 1992.

[15] Tziouvaras, D.A., G. Benmmouyal and J. Roberts, New multi-ended fault location design for two- or

three-terminal lines, Proc. 7th Int. IEE Conf. on Dev. in Power System Protection, pp. 395398,April 2001.

[16] Lin, Y., C. Yu and C. Liu, A New Fault Locator for Three Terminal Transm. Lines Using Two

Terminal Sync. Volt. And Cur. Phasors, IEEE Transactions on Power Delivery, Vol. 7, No.3, pp.

452459, July 2002.

[17] R. Syahputra, Fuzzy Multiple Objective App. for the Improvement of Distribution Net. Efficiency by

Considering Distributed Generation, IJCSIT, Vol 4, No 2, April 2012.


15/15


37

[18] Brahma, S.M., Fault Location Scheme for a Multi Terminal Transm. Line Using Synch. Voltage

Measurements, IEEE Transactions on Power Delivery, Vol. 20, No. 2, pp. 13251331, April 2005.

[19] Jang, J.S.R., 1993, "ANFIS: Adaptive-Network-based Fuzzy Inference System", IEEE Trans. Syst.,

Man, Cybern., 23, 665-685, June.

[20] B. R. Oswald and A. Panosyan, A New Method for the Computation of Faults on TransmissionLines, IEEE PES Transmission and Distribution Conference and Exposition Latin America,

Venezuela, 2006.[21] Izykowski, J., M.M. Saha, E. Rosolowski, P. Balcerek and M. Fulczyk, A Fault Location Method for

the Application with Current Dif. Relays of Three Terminal Lines, IEEE Transactions on Power

Delivery, Vol. 22, No. 4, pp. 20992107, October 2007.

[22] M.S. Sachdev and R. Agarwal., A Technique for Estimating Overhead Transmission Line Fault

Locations from Digital Impedance Relay Measurements, IEEE Transactions on Power Delivary, Vol.

3, No. 1, pp. 121-129, 1988.

AUTHOR

Ramadoni Syahputra received B.Sc. degree in Electrical Engineering from Institut

Teknologi Medan and M.Eng. degree from the Electrical Engineering Department,

Engineering Faculty, Gadjah Mada University, Yogyakarta, Indonesia, in 1998 and

2002, respectively. He was with the Electrical Engineering Department, Engineering

Faculty, Universitas Muhammadiyah Yogyakarta (UMY), Indonesia. His researchinterests include computational of power system, artificial intelligence in power system,

power system operation, power system control, power quality, distributed generation,

and renewable energy.

Date post:	03-Apr-2018
Category:	Documents
Upload:	anonymous-gl4irrjzn
View:	220 times
Download:	0 times

A Neuro-Fuzzy Approach for the Fault Location Estimation of Unsynchronized Two-Terminal Transmission...

Documents